[Pw_forum] a question about LDA+U

[Matteo Cococcioni]

Dear Ruijuan Xiao

I run the calculation of FeO in the smallest cell (the one with two Fe and 2 O)
and I attached the input and output to this mail.

It seems to me that the atomic coordinates of your input files are wrong. The
point, in my opinion, is that you decided to give them in alat units but in the
case of rombohedral cell the code fixes the coordinate axes in a strange way
(the z-axis is along the 111 direction of the cubic cell and the x and y axes
somewhere in the 111 plane). So the coordinate axes are not the edges of the
cubic cell anymore and it's constructing a crystal which is not the one you
have in mind. My suggestion would be to use always crystal corrdinates (though
it may require some initial conversion). You can realize this looking into my
output file.

Second issue. Even with correct atomic coordinates you still would not have got
the sixth electron localized in one state only if you are using the last
release (Espresso package). The reason for this, in my opinion, is that the
routine ns_adj is not able to suggest to the code which is the correct
electronic ground state (in this case the one with the sixth electron in the
Z^2 state which is along the 111 cubic diagonal). As Stefano Fabris was
explaining in his mail this is sometimes important (crucial for FeO) because
avoids that the code gets trapped in some local minima.
Attached to this mail you can also find an older version of the ns_adj routine
which you can insert in the new code. THe output I attach has been obtained
using this routine in an older version.

Remind that the ground state with last e- in the Z^2 orbital is not the true
ground state of FeO (see my thesis for details) but is for sure the one you
should obtain in the small cell.

I hope this can help you. 

Matteo




Quoting rjxiao <rjxiao at blem.ac.cn>:

> Dear Matteo Cococcioni,
> 
> Thank you for your reminding me to check the position of atoms. I have
> checked it carefully,but I am sorry that I can not agree with you.Does it
> because we choose different primitive cell? What I choose is the cell with
> lattic parameter of 5.30559 angstrom, and the angle between two lattic vector
> is 33.5573 degree. How about the primitive cell that you have chosen? If they
> are different, I can try that one.I am really thank you because I learn much
> from your thesie.
> 
> 
> Best regards,
> 
> Sincerely,
> Ruijuan Xiao
> Institute of Physics,
> Chinese Academy of Sciences
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
> 



-------------- next part --------------
A non-text attachment was scrubbed...
Name: FeO_in_out.tar.gz
Type: application/x-gzip
Size: 9409 bytes
Desc: not available
Url : /pipermail/attachments/20041023/740d8f1d/attachment.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ns_adj.f90
Type: application/octet-stream
Size: 4092 bytes
Desc: not available
Url : /pipermail/attachments/20041023/740d8f1d/attachment.obj